The horrific, caught-on-tape death of Eric Garner demands that America address the problems of its policing and criminal justice system, and the injustice, both racial and social, that clearly underpins these systems. Liberals and conservatives alike recognize this.

But we seem to be low on solutions.

After all, from a conservative perspective, it remains the case that crime, though at historic lows, is still bad, unavoidable, and requires police forces that will need to use violence and, human nature being what it is, use it wrong in some cases. And from a left-wing perspective, it can sometimes seem that systemic injustices are so deeply ingrained that no policy agenda can ever hope to fix them.

However, there is a solution. It's just not a national one.

One of the biggest structural problems in American culture today is the belief that national problems require national solutions; this has created the figure of the president as thaumaturgic king: half-Oprah, half-emperor. Our Washington overlords, the thinking goes, need to stand on the "right side of history" and craft grand solutions to our historic problems.

Instead, we should recognize that while we may not completely reverse racial and social injustice this side of the Eschaton, there are many local and practical things we can do to ensure that policing is both more respectful and still effective. After all, as everyone knows, the best thing for criminals is for local communities to stop trusting the police, which makes it in everyone's interest that policing be seen as fair.

We have all seen the astonishing violence and incompetence of the police in Ferguson. By contrast, few people read a story published in The New York Times shortly before the events of Ferguson, on the city of Camden, New Jersey. Camden was or could have been a Ferguson. But it replaced its police force, and by implementing smarter policing strategies in just two short years, got an astonishing drop in crime, as well as better community relations. What made Camden happen was not some national policy, it was social change at the local level.

Similarly, perhaps the most underreported story in America is what the Washington Monthly called the conservative war on prisons. At the national level, the rhetoric of the conservative movement is still pretty much tough on crime, full stop. But at the level of dozens of state legislatures, Christian conservatives, with fiscal conservatives now following in the wake of state budget crises, have led the charge for the kinds of prison reforms that progressives called for and never could get implemented. This is why deep-red Texas is now the model for smart prison reform.

Another example: Project HOPE, a Hawaii probation program that has shown striking results by relying on light, but swift and certain, punishment for misconduct instead of draconian but random punishment.

There are two key lessons to be learned from these examples.

The first, as I have said, is that keeping streets safe and reducing the number of future Eric Garners will require social change at the local level — one city hall, one state legislature, one police union at a time — driven by coalitions of progressives and Christian conservatives, as well as tough-minded-but-pragmatic law-and-order conservatives.

The second lesson is that there is actually very little we know about what works, and this is the other reason why we should go local. Most of the way we do social science is unscientific. The gold standard for evidence in social science is repeated randomized field trials, but we do very little of them, preferring to rely on the much blunter, biased instrument of statistical analysis.

The author Jim Manzi has painstakingly recorded all of the randomized field trials done in criminology over the past five decades, in the most important article one can read about social science. The total number is barely over 120. By contrast, a single large corporation will run thousands of such experiments every year, just to know what kind of coupon to send you. But we can't do research on a similar scale to figure out what will save lives from the destruction and ruin that our criminal justice system so easily metes out.

As Manzi points out, the vast majority of these social science experiments teach us nothing — but this is precisely the reason to amp up their numbers dramatically, because only large numbers of repeated experiments yield reliable social science insights. In fact, even these few experiments yielded the insights that led to the "Broken Windows" style of policing, which is credited with the astonishing crime drops of cities like New York. But now we are finding out that "Broken Windows" is far from perfect, and may not always work. We need to do thousands of criminology experiments per year to really understand what works and what doesn't. And those experiments can and must be done locally.

Surely there are policing and probationary methods and other crime policies that can deliver both low crime and much less victimization of minority and/or low-income communities. Camden is no utopia, but it seems to be a lot better than Ferguson.

But by and large, we have only hints as to what these policies are, though we do know they are extremely context sensitive.

We need a broad movement toward experimentation on policing and criminal justice policies at the local level, in concert with local actors.

I'm not a fan, to say the least, of the way Bill Gates single-handedly hijacked America's public education system by working local legislatures behind the scenes. But I have to say, I admire his style. It's not hard to think that a similarly smart, patient, and deep-pocketed philanthropist could also fund an effort to turn as many Fergusons into Camdens as possible, and to run as many criminology experiments as possible and, bit by bit, improve our system.

It would be a decades-long, decidedly unglamorous, decidedly non-headline-grabbing process. But it would save and transform countless lives and significantly ameliorate racial injustice in the process.